Components of primary care interventions to treat childhood overweight and obesity: a systematic review of effect


GM Sargent, Rural Clinical School, Faculty of Medicine, The University of New South Wales, PO Box 5695, Wagga Wagga, NSW 2650, Australia. E-mail:


The primary care setting presents an opportunity for intervention of overweight and obese children but is in need of a feasible model-of-care with demonstrated effectiveness. The aims were to (i) identify controlled interventions that treated childhood overweight or obesity in either a primary care setting or with the involvement of a primary healthcare professional and (ii) examine components of those interventions associated with effective outcomes in order to inform future intervention trials in primary care settings. Major health and medicine databases were searched: MEDLINE, CINAHL, EMBASE, Cochrane Reviews, CENTRAL, DARE, PsychINFO and ERIC. Articles were excluded if they described primary prevention interventions, involved surgical or pharmacological treatment, were published before 1990 or not published in English. Twenty-two papers describing 17 studies were included. Twelve studies reported at least one significant intervention effect. Comparison of these 12 interventions provides evidence for: training for health professionals before intervention delivery; behaviour change options (including healthy diet, activity and sedentary behaviour); effecting behaviour change via a combination of counselling, education, written resources, support and motivation; and tailoring intensity according to whether behavioural, anthropometric or metabolic changes are the priority. These components are practicable to future intervention studies in primary care.


It has been predicted that due to the health consequences and prevalence of obesity, the current generation of children may have a shorter life expectancy than their parents (1). Although prevention is important (2), established overweight and obesity in children is unlikely to resolve without active intervention (3).

A recent Cochrane systematic review of interventions for treating obesity in children (4) found positive effects in several studies. However, there was little high-quality evidence to recommend one treatment over another and the review raised the question as to what methods work best in different healthcare settings (4). Intervention research to date has focused on community settings, particularly schools, and the resource intensive tertiary healthcare setting (5).

The primary care setting has the potential to access a large proportion of the community throughout the life course and presents an intervention opportunity for overweight and obese children (6). Systematic reviews of the treatment of childhood obesity have not previously focused on the question of effectiveness of interventions through primary care. Primary care is in need of a feasible model-of-care with demonstrated effectiveness in the treatment of overweight or obese children.

To account for international variations in terminology, primary care refers to first-contact health care services located in the community that are directly accessible to the public. In accordance with this, a variety of health professionals may have a role in primary care delivery including medical, nursing and allied health professionals (7) (hereafter referred to as primary health care professionals).

The first aim of this systematic review was to identify controlled interventions that treated childhood overweight or obesity in either a primary care setting or with the involvement of a primary healthcare professional. The second aim was to examine components of those interventions associated with effective outcomes in order to inform future intervention trials in primary care settings.


This systematic review was conducted and reported in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (8).

Key questions

The two key questions that guided the literature search strategy were (i) What does the existing literature report on effectiveness of interventions involving primary care in the treatment of overweight or obese children? and (ii) What components of these interventions are associated with significant outcomes?

Information sources

Major health and medicine databases of published literature were searched between September and November 2007. These were: MEDLINE, EMBASE, CINAHL, Cochrane Reviews, CENTRAL, DARE, PsychINFO and ERIC. The bibliographies of included articles were hand searched and promising titles reviewed in order to locate articles not catalogued in the major databases.

Search strategy

The database search strategy (Table 1) was constructed using medical subject headings, and keywords combined into three groups defining: the target population (overweight and obese), the type of intervention (including treatment, therapy, intervention, management assessment, delivery) and the desired setting (including general practice, family practice, primary care, medical staff, nursing staff, physician's office). Articles retrieved by the search strategy had at least one term from each grouping. Keywords in the title field identified articles in which treatment of childhood overweight and obesity was not the focus and those reporting surgical or pharmacotherapy interventions and removed these from the search results. Word truncation and wildcards allowed for variations in spelling and word endings. The strategy was adapted minimally between databases to allow for differences in accepted search terms and limits. Database limits for the year of publication and English full text were applied if available. No search limits were set on research or publication type.

Table 1.  Medline (Ovid) search strategy
1. exp body weight changes/or exp overweight/or exp obesity/
2. (overweight or obese or obesity).mp.
3. 1 and 2
4. (diabet$ or asthma$ or prader-willi).ti.
5. (neoplasm$ or cancer$).ti.
6. (bariatr$ or bypass$ or intragastric or gastric band$ or laparoscopic or surgery).ti.
7. (schizoph$ or psychiatric patient$ or psycho$ or lamotrigine).ti.
8. (orlistat or sibutramine or rimonobant).ti.
9. exp Apnea/or sleep apnea.ti.
10. 4 or 5 or 6 or 7 or 8 or 9
11. 3 not 10
12. (assess$ or impact or impact asess$ or evaluat$ or effect$ or evidence or compar$).ti.
13. (intervention$ or investigat$ or result$).ti.
14. prevent$.ti.
15. (manag$ or treat$ or therap$ or deliver$ or care).ti.
16. (perception$ or belief$ or barrier$ or understand$ or qualitative or focus group or role or roles).ti.
17. 12 or 13 or 14 or 15 or 16
18. (nurse$ or physician$ or general practi$ or primary care or referr$).ti.
19. 17 or 18
20. 11 and 19
21. (primary care or ambulatory care).mp.
22. (family practi$ or general practi$).mp.
23. (physician$ or practitioner$).mp.
24. (practice nurse$ or school nurse$ or nurse practitioner$).mp.
25. multidisciplinary
26. (dietitian$ or psychologist$ or clinician$).mp.
27. ((health or health-care or healthcare) adj (professional$ or personnel or profession$)).mp.
28. exp comprehensive health care/or exp primary health care/or exp critical pathways/or exp patient care team/or exp patient-centered care/
29. allied health personnel/or community health aides/or physician assistants/or pediatric assistants/or exp health educators/or medical staff/or exp nurse clinicians/or exp nurse practitioners/or nursing staff/or exp physicians, family/
30. 21 or 22 or 23 or 24 or 25 or 26 or 27 or 28 or 29
31. 20 and 30
32. limit 31 to ‘all child (0 to 18 years)’
33. limit 32 to english language
34. limit 33 to yr = ‘1990–2007’

Eligibility criteria

Articles were included if they described an intervention trial that aimed to treat infants, children or adolescents that were overweight or obese. Articles describing both randomized and non-randomized controlled intervention trials (RCTs and non-RCTs) were eligible for inclusion.

Articles were excluded if they described primary prevention interventions or if the majority of participants were over 18 years of age. Articles published in a language other than English, or before 1990 were excluded as were articles where participants had undergone surgery or if obesity was due to pharmacotherapy or a congenital disorder. Those that did not clearly describe the involvement of either a primary care setting or primary healthcare professional were excluded. No restrictions were placed on the outcome measures being reported.

Study selection

An initial trial of the inclusion and exclusion criteria by two reviewers resulted in 78% reviewer concordance (the proportion of titles on which the two reviewers agreed to include/exclude the first 543 titles). The eligibility criteria were refined and titles reviewed a second time, which resulted in 99% concordance. One reviewer completed the review of the remaining titles and abstracts (first screening stage, Fig. 1). The full-text articles were reviewed in order to identify controlled trials (second screening stage). Any that were equivocal were submitted to a second reviewer and included/excluded accordingly. Two reviewers examined the full articles in order to identify those with explicit involvement of either a primary care setting or primary healthcare professionals (eligibility stage).

Figure 1.

Flow of papers through selection process, using Preferred Reporting Items for Systematic Reviews and Meta-Analyses format (8).

Data collection process

Data from included studies were extracted by one reviewer (GS) into a purpose-built database using a framework adapted from the Effective Practice and Organisation of Care data collection checklist (9). The database captured details covering: study design and implementation, target participants; group treatment; process measures and outcomes. Data describing interventions that were reported in more than one article were extracted together. Following completion of the data extraction process, data from the first four studies were extracted a second time in order to ensure consistency.

Synthesis of results

Effective interventions were regarded as those with a statistically significant improvement (P < 0.05) in any outcome measure when compared with the control or comparison group at the first post-intervention follow-up. Each outcome measure was categorized into one of four domains: anthropometric/body composition, metabolic, behaviour change or psychosocial. Significant intervention outcomes were cross-referenced to intervention components for comparison of effect. It was considered inappropriate to compare intervention components between studies that reported no significant outcomes as this lack of significant effect may be attributable to variables other than intervention components. Intervention groups from studies with more than one treatment intervention arm were treated as separate interventions for the purposes of examining varied components.

Data items – intervention components

Detailed examination of the following components of effective interventions were conducted: intervention setting, the involvement of primary healthcare professionals in intervention delivery, personnel training as part of the intervention, behaviour change targets, mode used to affect behaviour change, role of parents and child as agent-of-change, number of contacts, rate of contacts and treatment length. Components were categorized, when appropriate, in order to facilitate comparison of intervention components (Table 2).

Table 2.  Description of the categories used for data analysis of intervention components
ComponentCategories (description)
  • Categories were defined according to included studies.

  • *

    If interventions involved more than one setting the setting was considered to be that where the majority of contacts with a health professional had taken place. Some clarification of country of origin specific terminology was necessary in order to classify these accordingly.

  • Multiple categories allowed.

  • Intervention contacts occurred when a PHCP was scheduled to either meet or communicate (by telephone or mail) with a participant (child or parent). Visits clearly for assessment purposes only were not included. The average rate of contacts per month (rounded to one decimal place) was calculated by dividing the recommended number of contacts by the intervention delivery period in months.

  • PHCP, primary healthcare professional.

Intervention setting*1. General or family practice (included family paediatrician office);
2. Health centre (as described in publication);
3. Other primary care (if setting described as primary care but not explicit);
4. School (only included if delivery involved PHCP);
5. Hospital out-patient clinic (only included if delivery included contact with PHCP).
Primary healthcare professional1. Medical (general practitioner, family practitioner, family doctor, medical doctor, school doctor, family paediatrician, primary care physician);
2. Nursing (primary care nurse, school nurse, practice nurse, nurse);
3. Allied health (dietitian, nutritionist, exercise physiologist, psychologist and counsellors).
Behaviour change targets1. Maintaining a calorie restricted diet;
2. Low fat or low sugar diet;
3. Achieving a healthier diet (other than 1 or 2);
4. Attending physical activity sessions;
5. Increasing lifestyle activity;
6. Decreasing sedentary behaviour;
7. Incorporating both healthier diet and activity into the daily routine (healthy lifestyle).
Strategies to achieve behaviour change1. Motivation or support (including improvement of self-efficacy and incentives);
2. Counselling or education (including behaviour modification, parenting skills, coping skills and increasing knowledge of childhood overweight and obesity);
3. Provision of written resources;
4. Provision of physical activity sessions;
5. Physical activity and dietary prescription.
Agent-of-change1. The child was considered sole agent-of-change (parent gave consent for child to participate and may have accompanied the child to some sessions, but no responsibility for change was given to the parent).
2. The parent and child were joint agents-of-change (parent and child attended at least some sessions independently but sole responsibility for change was not given to parents).
3. Parents were sole agents-of-change when given express responsibility for change (sessions may have been attended with or without the child).
Intensity of contact rate1. Less than one contact per month (rate <1);
2. Monthly contacts (rate 1–2);
3. Fortnightly or weekly contacts (rate 2–4);
4. More than one contact per week (rate >4).

Risk of bias

Criteria for assessing the methodological rigour (MR) of included articles were adapted from those in use by the Cochrane Effective Practice and Organisation of Care Review Group (9) and recent systematic reviews (5,10). The 10 criteria were scored objectively using published data and reflect potential bias in: participant allocation, participant attrition/exclusion, measurement, analysis and reporting of outcomes and indicate rigour of implementation. The results were itemized for each study and a composite MR score presented in outcome data tables in order to provide a rapid indication of study rigour without excluding potentially useful studies.


Study selection

The database search strategy (Table 1) identified 956 articles for the review process. The selection process (Fig. 1) resulted in 22 articles (11–32) eligible for inclusion, seven from Medline, 12 from EMBASE and three through bibliography and citation searches. These articles reported on 17 separate studies (hereafter referred to by the first post-intervention outcomes paper).

Methodological rigour score

The MR score varied between two and nine out of a possible score of 10 (Table 3). Although all studies included in the review were controlled intervention studies, three studies did not establish an effective control group and reported no between-group comparisons (13,20,21). Seven studies reported having reached the calculated effective sample size (16–18,22,27,29,31). Seven studies reported conducting intention-to-treat analyses (17,18,24,27,29,31,32). There were few articles with a high MR.

Table 3.  Methodological rigour (MR) of included studies (structured by year of publication)
First author and citation(s)RandomizationBlindingParticipant inclusion/exclusion criteria adequately describedIntent-to-treat analysis performedEffective sample size calculations shownEffective control groupStandard measures describedComparison of baseline parameters of completers to non-completersOutcome data for ≥80% of participantsStakeholder involvement in designMR score/10
  • *

    Intervention reported significant effect.

  • Indicates studies that demonstrated an effective sample size was reached.

  • Indicates studies that stated effective sample size reached, but data not presented (no score).

  • §

    Indicates effective sample size was calculated but not reached.

Nuutinen, 1991 (11,12)NoNoNoNoNoYesYesNoYesNo3*
Israel, 1994 (13)YesNoNoNoNoNoNoYesNoNo2
Muller, 2001 (14–16)NoYesYesNoYesYesNoYesNoNo5*
Nova, 2001 (17)YesNoNoYesNoYesNoYesNoNo4*
Saelens, 2002 (18)YesNoNoYesYesYesYesNoYesNo6*
Eliakim, 2004 (19)NoNoNoNoNoYesYesNoNoYes3*
Korsten-Reck, 2005 (20)NoNoYesNoNoNoYesNoYesNo3
Lane-Tillerson, 2005 (21)NoNoNoNoNoNoYesNoYesNo2
Nemet, 2005 (22)YesNoYesNoYesYesYesNoYesNo6*
Graf, 2006 (23)NoNoNoNoNoYesYesNoYesNo3*
Patrick, 2006 (24,25)YesNoYesYesNoYesYesYesYesNo7*
Gillis, 2007 (26)YesNoNoNoNoYesYesNoNoNo3
Golley, 2007 (27)YesYesYesYesYesYesYesYesYesNo9
Kalavainen, 2007 (28)YesNoYesNoYes§YesYesNoYesNo6*
McCallum, 2007 (29,30)YesYesYesYesYesYesYesYesYesNo9*
Savoye, 2007 (31)YesNoYesYesYesYesYesNoNoNo6*
Schwartz, 2007 (32)NoNoYesYesNoYesYesYesNoNo5*

Study characteristics

Study characteristics are summarized in Table 4. Studies were carried out in six different countries: USA (six), Germany (three), Israel (three), Australia (two), Finland (two) and Italy (one). Of the 22 articles, three (11–13) were published in the 1990s and the remainder in or after 2000.

Table 4.  Summary of 17 included studies (structured by year of publication)
First author and citation(s) (country and programme name)Study design (control type)Brief description of intervention including primary care involvementStudy population: age range (actual mean and distribution when available), body composition inclusion criteria, risk level*Sample size (% male)MR score/10
  • *

    Study population risk level terminology standardized using the following definitions: normal weight, BMI 5th–85th percentile for age and sex; overweight, BMI 85th–95th percentile for age and sex; obese, BMI ≥95th percentile for age and sex or weight ≥120% of average weight-for-height; at risk of overweight, <85th BMI percentile with at least one parent with BMI >30.

  • Number of participants in either intervention or control groups for which baseline data were reported.

  • Intervention with significantly effective outcomes.

  • §

    Indicates studies that demonstrated an effective sample size was reached.

  • Indicates studies that stated effective sample size reached, but data not presented.

  • **

    Indicates effective sample size was calculated but not reached.

  • BMI, body mass index; CDC, Centers for Disease Control; CDC&P, Centers for Disease Control and Prevention; FITOC, Freiburg Intervention Trial for Obese Children; IOTF, International Obesity Task Force; KOPS, Kiel Obesity Prevention Study; LEAP, Live, Eat and Play; MR, methodological rigour; non-RCT, non-randomized controlled trial; NS, data not stated; PACE+, Patient-Centered Assessment and Counseling for Exercise plus Nutrition; RCT, randomized controlled trial; SD, standard deviation.

Nuutinen, 1991 (11,12)
Non-RCT (usual care and normal weight controls)Intensive dietary counselling to affect change in dietary intake. Parent and child encouraged to make informed decisions. Monthly physician visits, plus five to seven education sessions with nutritionist/psychologist in health centre as individual families (30 min) or small group (60 min). Parent attended all sessions, child attendance varied according to child's age. Usual care control consisted of monthly meetings between child and nurse at school.6–16 years old (mean 11.3). Relative body weight > 120% (using Finnish data). Obese*77 (74% male)3
Israel, 1994 (13)
RCT (alternate intervention group)Comparison of effect of agent-of-change: Group 1 parents given responsibility for motivating children to change; Group 2 emphasis on child control with parent support. Recruitment by school nurses, paediatricians or via newspaper articles. Delivery setting not clear. Parent and child groups separate, consisted of 17 education sessions (8 × weekly, 9 × fortnightly) covering cue control, physical activity, food intake, rewards and parent training in child-management skills.8–13 years old (mean 10 years 11 months, SD 1.2; range 8 years 11 months to 13.0 years). Weight for height > 20% above norms for age and sex (population not specified). Obese*34 (NS)2
Muller, 2001 (14–16)
(Germany, KOPS)
Non-RCT (no treatment control)School-based nutritional and activity education programme delivered principally by nutritionist, with class teacher (following nutrition education training) assisting. Initial assessment by physician. Additional counselling sessions by a nutritionist targeted families with an overweight or obese child or an overweight parent. Four key messages: eat fruit and vegetables every day, reduce intake of high fat food, keep active for at least 1 h per day and limit TV to less than 1 h per day.First year at school (5–7 years old). Triceps skin-fold ≥90th percentile (using German reference data). At risk of overweight (overweight parent), overweight and obese*297§ (48.8% male)5
Nova, 2001 (17)
RCT, randomised at physician level (usual care control)A brief office-based intervention delivered by family paediatrician. Aim to assess the effect of two different levels of intensity and information on parental commitment, family compliance, attendance and child weight loss. Child and parent attended five counselling sessions with family paediatrician describing 1400 calorie diet, physical activity requirements and provision of written resources. Parents attended all visits and were expected to be committed to the programme at home. Paediatrician received one session of training plus written updates.3–12 years old (mean 8.6 ± 2). Weight ≥ 20% over 50th percentile for height (using Eid Index). Obese*186 (56% male)4
Saelens, 2002 (18)
(USA, Healthy Habits plus PACE+)
RCT (usual care control)Telephone- and mail-based intervention to evaluate effectiveness, patient satisfaction and acceptability of a behavioural weight control programme initiated in primary care paediatric clinic. Computer tailored behaviour change targets and strategies (physical activity, sedentary behaviour, dietary fat intake or fruit and vegetable intake). Two visits with primary care paediatrician (one with parent) plus weekly/biweekly telephone counselling (with psychologist or nutritionist) and mailed resources for child. All providers attended training before intervention.12–16 years old (mean 14.2, SD 1.2). BMI 20–100% above 50th percentile for sex and age (using US population data [Rosner et al., J pediatrics 1998;132(2):211–222]. Overweight and obese*44§ (59% male)6
Eliakim, 2004 (19)
Non-RCT (usual care control)Educational intervention aimed at initiation and long-term maintenance of lifestyle changes (combined dietary, behavioural, physical activity) for families seeking an intervention longer than initial 3 months (22). Parent/s and child attended physician counselling session and four education sessions (with physician and dietitian) in child health and sports centre. Child/parents attended six visits with dietitian (with or without child according to age) covering nutrition education and intake of hypocaloric diet (15%–30% less than baseline intake). Child attended physical activity twice weekly in health centre. Psychologist involved with intervention design and implementation.6–16 years old (NS). Target not clearly defined (mean BMI percentile 96.5 using USA CDC growth charts). Obese114 (NS)3
Korsten-Reck, 2005 (20)
(Germany, FITOC)
Non-RCT (no treatment control)Intensive intervention consisting of regular physical exercise, dietary and behavioural education and goal-setting to achieve long-term changes in lifestyle behavioural patterns and self-esteem. Delivered by interdisciplinary team in hospital out-patient clinic following participant recruitment in primary care by family doctors, paediatricians or school doctors. Parent and child attended one physician visit together, monthly education sessions (separately) plus physical activity (child only) three times per week.9–12 years old (mean 10.5 ± 1.5). BMI >97th percentile, or 90th–97th with a somatic comorbidity or one overweight parent (using German reference data). Overweight and obese*531 (46% male)3
Lane-Tillerson, 2005 (21)
Non-RCT (three alternate delivery formats)Programme conducted exclusively by healthcare professionals stressing the importance of healthy diet, physical activity, behaviour modification and use of goal setting to attain and maintain changes. Sixteen groups sessions with a nurse delivered in health centre in three group formats: daughter alone, daughter and mother together and daughter and mother separately.13–17 years old (mean 15). Adolescent African–American girls BMI ≥85th percentile for age chart (using USA CDC growth charts). Overweight and obese*18 (0% male)2
Nemet, 2005 (22)
RCT (usual care control)Combined dietary/behavioural/physical activity intervention with aim to engage whole family in the ‘battle against overweight’, examines short- and long-term effects on physical and behavioural parameters. Parent/s and child attended four education sessions (with physician and dietitian) in health centre setting. Parents attended six visits with dietitian (with or without child according to age) covered intake of hypocaloric diet (less 15 to 30%) and nutrition education. Child attended 1-h physical activity sessions delivered by ‘physicians’ (ex-track and field athletes) twice weekly in health centre.6–16 years old (control: 11.3 ± 2.8, intervention: 10.9 ± 1.9). BMI >95th percentile for age and sex (mean BMI: control 27.8 ± 5.0, intervention 28.5 ± 4.1, using USA CDC growth charts). Obese*54§ (57% male)6
Graf, 2006 (23)
(Germany, StEP TWO)
Non-RCT (no treatment and non-participant controls)Twice weekly after school intervention. Children cooked and ate with nutritionist then attended health education with medical doctors (human body, anatomy, consequences of inactivity, digestion, energy supply and use) followed by physical education (60–90 min) with gymnasts (60 sessions at school). Parents attended six education sessions (parents only) and two activity days (with child).6–11 years old (range 5.8–11.7). BMI >90th percentile (using German reference data). Overweight and obese*276 (NS)3
Patrick, 2006 (24,25)
RCT (alternate sun protection intervention)Telephone- and mail-based intervention initiated in primary care with computer tailored behaviour change targets (physical activity, sedentary behaviour, dietary fat intake or fruit and vegetable intake). Primary care providers from private clinics undertook 2 h of training and delivered one 3–5 min visit. Monthly 10–15 min telephone counselling and written resources were supplied and mailed. Focus on adolescent as sole agent-of-change11–15 years old (mean 12.7, SD 1.3). General population (results analysed according to BMI percentiles <85th, 85th–95th and >95th (using USA CDC growth charts). Normal weight, overweight and obese819 (46.5% male)7
Gillis, 2007 (26)
RCT (waitlist control)Minimal health education intervention with aim to decrease weight and adverse metabolic consequences of obesity. Delivered by a single practitioner in primary care setting. Three counselling visits with physician, plus weekly telephone counselling on maintaining a healthy lifestyle (diet and exercise) and use of resources (all child only).7–16 years old (NS). Ultra-orthodox Jewish children (minimal exposure to media). BMI >90th percentile (reference population not specified). Overweight and obese*27 (NS)3
Golley, 2007 (27)
RCT (with assessor-blinding) (waitlist control)Comparison of effectiveness of parenting skills alone (group 1) vs. parenting skills plus healthy lifestyle education (group 2), both using a modified Triple P parenting programme to improve parent's efficacy to affect behaviour modification. Initial assessment by medical practitioner, parent education with dietitian in hospital out-patient clinic. Dietitian trained in use of Triple P. Group 1: four education sessions (weekly), plus telephone contacts (4 × weekly plus 3 × monthly). Group 2: 11 education sessions (4 × weekly, 4 × fortnightly, 3 × monthly).6–9 years old (NS). Overweight according to IOTF definition [Cole et al., BMJ 2000; 320:1240–3], excluded if BMI z-score >3.5. Overweight and obese*111§ (37% male)9
Kalavainen, 2007 (28)
RCT (usual care control)Family-based behavioural and solution-oriented therapy with motivation support. Focused on using nutrition and exercise to maintain long-term healthy lifestyle and well-being of child. Initial recruitment, baseline assessment and control group delivered in health centre by school nurse. Parent and child (separately) attended five education plus activity sessions in out-patient clinic. Dietitian delivered group intervention sessions to parents. Student nutritionists led child group sessions (after receiving training from dietitian). Parents expected to be agent-of-change for whole family.7–9 years old (mean 8.1, SD 0.8, range 6.6–9.7). Weight for height of 120–200% (mean 142%, range 115–182%, using Finnish national growth charts). Obese*70** (40% male)6
McCallum, 2007 (29,30)
(Australia, LEAP)
RCT (usual care control)Brief solution-focused family therapy with a focus on behaviour change to achieve healthy lifestyle goals using a tailored information package. Four standard length consults (parent and child) with general practitioner in general practice setting (delivered after general practitioner attended 5–7.5 h of training).5.0–9.11 years old (NS). Overweight and obese according to IOTF cut-points Cole et al. 2000, excluded if BMI z-score ≥3.0 (UK reference data).Overweight and obese*163§ (NS)9
Savoye, 2007 (31)
(USA, Bright Bodies)
RCT (waitlist control)Intense, family-based, non-diet programme focused on nutrition, exercise and behaviour change in order to decrease health effects in adulthood. Initial recruitment and referral by a physician. Weekly nutrition/behaviour modification skills and physical activity twice/week for 6 months then fortnightly for 6 months (dietitian, social worker and exercise physiologist in hospital out-patient clinic). Counsellors and therapist team received specific training. Parents attended all sessions with child and asked to model health behaviour change.8–16 years old (NS). Inner-city minority children BMI >95th percentile (mean BMI 36, using USA CDC&P growth charts). Obese*174§ (39% male)6
Schwartz, 2007 (32)
Non-RCT (usual care control)Parents attended motivational interviewing session/s in primary care setting. Group 1: attended one 10–15 min motivational interviewing session with paediatrician. Group 2: attended two motivational interviewing sessions with each a paediatrician (10–15 min) and dietitian (45–50 min). Paediatrician and dietitian attended 2-day training before intervention. Parents given information sheets on healthy eating and activity and a video on parent behaviour around feeding issues.3–7 years old (mean 5.3 years, control mean 4.7). BMI 85th–95th percentile (using USA CDC growth charts) or parent with BMI ≥ 30. Overweight and at risk of overweight*91 (41% male)5

The 17 studies consisted of 10 RCTs and seven non-RCTs. Randomization was implemented at the participant level in nine studies (before and after baseline assessment) and at the physician level in one (17). The randomization methods described were computer generated numbers (22,29) and envelopes (27,28). Various criteria were used to stratify before randomization, namely: overweight status (13,29), delivery site (29), recruitment site (27), scheduling constraints (13), child age (13) and sex (13,27). Three studies reported that data collection was carried out by assessors who were blinded to participant group allocation (14,27,29).

Seven interventions targeted obese children exclusively. Nine interventions targeted overweight and obese children, two of these included children who were normal weight or considered at risk of overweight due to one obese parent. One intervention focused on overweight children and included normal weight children considered at risk due to parental obesity.

The age of children targeted for intervention participation ranged from 3 to 17 years old; however, the actual age distribution of children participating in studies was not consistently reported. Five studies did not report the sex distribution of participants and one intervention involved female participants only (21). The other 11 studies had an average of 49.6% (median 46.5%) male participants. Many interventions targeted a large age range but had insufficient sample size to control for age and sex.

The sample size for the intervention trials ranged from 18 to 819 with a median of 111 participants per intervention and 36 per group. The number of intervention groups and structure of control groups varied considerably between studies, particularly among non-RCTs. Control groups received from zero (no treatment) to 12 contacts. The terminology used to describe these does not necessarily reflect the dosage; for example, the ‘standard treatment’ control described in the study by Israel et al. (13) consisted of a very similar intervention to the alternate intervention being trialled.

Intervention evaluation

The outcomes reported by each study are illustrated by shaded squares in Table 5. The 17 studies each evaluated between four and 15 outcome measures, with a combined total of 30 different outcomes. All studies evaluated anthropometric outcomes, nine also measured metabolic outcomes, 10 evaluated behavioural change and eight assessed psychosocial change. Only one study included outcome measures from all four domains (22). The majority of behaviour change and psychosocial outcomes relied on self-report measures. An objective measure (accelerometry) to verify self-reported changes in sedentary behaviour and physical activity was used in one study (24). Interestingly, only the self-report (7-day activity recall) measure indicated significant between-group change. Results pertaining to the cost of the intervention were presented in one study (29).

Table 5.  Post-intervention outcome measures, by study (structured by year of publication) Thumbnail image of

Synthesis of significant outcomes

The outcome measures that were significantly affected by the intervention are indicated by a tick in Table 5. Statistically significant outcomes, when compared with a control/comparison group at the first post-intervention assessment, were reported in 12 of the 17 included studies. Eight of the 17 studies (47%) reported significant anthropometric changes. Of the nine studies that included metabolic measures, three (30%) reported significant outcomes. Of the 10 studies that included measures of behaviour, six (60%) reported significant changes. No significant psychosocial changes or other adverse effects were recorded. Meta-analysis of effect size was not possible due to the heterogeneity of outcome measures.

Significant outcomes from each study immediately after intervention and at subsequent follow-up are summarized in Table 6. Nine studies reported subsequent follow-up, which took place between 6 months and 4 years after initial post-intervention measures. Seven of the nine studies reported effective outcomes immediately after intervention and these all maintained some significant effect at subsequent follow-up.

Table 6.  Summary of significant intervention outcomes (structured by year of publication)
First author and citation(s)Methodological rigour scoreSample sizeIntervention group sample sizeIntervention group completion rate (%)Intervention duration (number of contacts)Contact rate (per month)*Significant outcomes (to control group) at initial post-intervention follow-upSubsequent follow-up period after interventionSignificant outcomes at subsequent follow-up
  • *

    Total number of contacts divided by the length of intervention delivery period in months. Referred to in text by four categories of intensity: less than one contact per month (rate <1), monthly contacts (rate 1–2), fortnightly or weekly contacts (rate 2–4) or more than one contact per week (rate >4).

  • Intervention with significant effect, significant outcome measures are in bold type.

  • Number of participants completing intervention.

  • BMI, body mass index; CI, confidence interval; HOMA-IR, homeostasis model assessment of insulin resistance; LDL, low-density lipoprotein; NS, data not stated.

Nuutinen, 1991 (11,12)377329112 months (17–19)1.4–1.6Dietary intake (lower sucrose consumption, P < 0.05).1 year and 4 yearsDietary intake (sucrose) −7 g d−1 vs. +9 g d−1) (P < 0.001).
Israel, 1994 (13)23418786 months (18)3Nil6 months and 2.5 yearsNil
Muller, 2001 (14–16),529713625Not clear, <1 school year (8–13 h)Not clearMedian triceps skin-fold increase was lower in intervention group (+11.3 mm vs. +13 mm, P < 0.01). Fat mass percentage (calculated from triceps skin-fold) increase was lower in intervention group (+0.4% vs. +3.6%, P < 0.05).4 yearsProportion of girls no longer considered overweight was 40.9% and 50% (as compared with 22.2% and 29.3 % for control, P < 0.034, measured by skin-fold and waist circumference respectively).
Nova, 2001 (17)418672716 months (5)0.8Percentage overweight decreased (−8.8% vs. −2.9%, P = 0.0001).6 monthsPercentage overweight decreased (−8.5% vs. −2.9%, P = 0.002)
Saelens, 2002 (18)644201004 months (16)4Proportion of children with a decreased BMI z-score was higher in intervention group (40% vs. 10.5% in control, P < 0.04). Eating and physical activity behavioural skills use was higher in intervention group (F > 5.17, P < 0.03 by self-report, F > 4.5, P < 0.04 by parent report).3 monthsProportion of participants with decreased BMI z-score (55.6% vs. 15.8%, P < 0.02).
Eliakim, 2004 (19)311477NS12 months (114)9.5BMI (−2.7 kg m−2, P < 0.05); BMI percentile (−5.2%, P < 0.05); and weight (−4.3 kg compared with control, P < 0.05)Nil 
Korsten-Reck, 2005 (20)3531496908 months (100)12.5Nil(intended at 12 months)Not yet published
Lane-Tillerson, 2005 (21)2186 (×3)1005 months (16)3.2NilNil 
Nemet, 2005 (22)65430803 months (34)11.3Body weight (−4 kg); BMI (−1.5 kg m−2); body fat percentage (calculated from skin-fold, −4.7%); total cholesterol (−24.2 mg dL−1); LDL (−14.8 mg dL−1); habitual physical activity index (+23.2 vs. −2.2 units); and endurance time (+161 vs. +45 s), all P < 0.05).12 monthsBody weight (+0.6 kg vs. +5.3 kg), BMI (−1.6 vs. +0.6), and body fat percentage (−2.3% vs. +3.5%), endurance time (+186 vs. 40 s), habitual physical activity index (+9.1 vs. −7.6), all P < 0.05.
Graf, 2006 (23)327646879 months (60)6.7BMI z-score decreased (1.99–1.83 vs. no treatment control 1.87–1.82, P = 0.028). Systolic blood pressure was significantly decreased in both intervention participants −9.5 mmHg, P = 0.002 and non-participants-5.3 mmHg, P = 0.025 vs. no-treatment control +0.5 mmHg).Nil 
Patrick, 2006 (24,25)78194248412 months (12)1Sedentary behaviour decreased (4.3–3.2 h d−1 vs. 4.2–4.3 h, P < 0.001). Only boys reported a significant increase in physical activity (4.1–4.4 days of moderate activity per week, P < 0.01 vs. no change from 3.8 days per week) as measured by 7-day recall, but this not reflected in accelerometry. Dietary intake (proportion meeting dietary guidelines for saturated fat intake) in girls (23.4% to 41% vs. 18.5–31%, 95% CI).Nil 
Gillis, 2007 (26)32714796 months (12–24)2–4NilNil 
Golley, 2007 (27)911137/38756 months (11)1.8Control group outcomes not measured6 monthsNil
Kalavainen, 2007 (28)6703597%6 months (16)4Weight for height (−6.8% vs. −1.8%, P = 0.001). BMI (−0.8 vs. 0.0, P = 0.003); and BMI z-score (−0.3 vs. −0.2, P = 0.022). A higher proportion of participants decreased weight-for-height by >5% (63% vs. 26%, P = 0.004)6 monthsWeight-for-height decrease of −3.4% (vs. +1.8%, P = 0.008), BMI increased less (+0.1 vs. +0.8, P = 0.016). Proportion of participants that decreased weight-for-height by >5% (44% vs. 17%, P = 0.015).
McCallum, 2007 (29,30)916382963 months (4)1.3Nutrition scores (dietary intake) measured using 4-day food record, +2.7 vs. +0.2 in control, P < 0.001), and was mostly attributable to replacement of full fat milk with low fat milk and water.12 months (variable)Nutrition scores (dietary intake) +2.4 vs. −0.2, P < 0.001).
Savoye, 2007 (31)61741057112 months (64)5.3BMI (−1.7 vs. +1.6, P < 0.001); estimated fat mass (−3.7 kg vs. +5.5 kg, P < 0.001); weight (+0.3 kg vs. +7.7 kg, P < 0.001); body fat (−4.0% vs. +2%, P < 0.001); total cholesterol (−9.2 mg dL–1 vs. +3.7 mg dL−1, P = 0.005), fasting insulin (−6.1 µIU mL−1 vs. +4.5 µIU mL−1, P < 0.001) and HOMA-IR (ratio of fasting blood glucose: insulin response, −1.52 vs. +0.90, P < 0.001).Nil 
Schwartz, 2007 (32)59140 Group 1
30 Group 2
68 Group 1
50 Group 2
6 months (1: Group 1, 4: Group 2)0.2: Group 1
0.7: Group 2
Child's intake of snacks decreased in Group 1 (compared with control, P = 0.01, parent report).
Decrease in dining out reported in Group 2 (compared with Group 1, P = 0.04).

Components of effective interventions

Intervention setting

Of the 17 studies included in this review, 11 interventions were principally delivered in primary care. The settings of interventions with significantly effective outcomes were: general or family practice (n = 2), health centre (n = 4), other primary care (n = 3), school (n = 2) and hospital out-patient clinic (after assessment by primary care physician, n = 1).

Involvement of primary healthcare professionals in intervention delivery

The two intervention arms described by Schwartz et al. (32) differed in the health professionals delivering the intervention; hence, these two groups were examined separately and 13 groups (from the 12 effective studies) were reported in these results.

Nurses were involved in participant assessment (11) and delivery of control conditions (28); however, none of the interventions that reported successful outcomes was delivered by a nurse. Hence, primary healthcare professional involvement in effective intervention arms comprised: delivery by medical professionals only (n = 3), delivery by both medical and allied professionals (n = 7) or delivery by allied health professionals after initial assessment and referral by medical and nursing staff (n = 3). Significant outcomes according to health professional involvement are summarized in Table 7.

Table 7.  Comparison of PHCP involvement in intervention delivery of each study with effective outcomes immediately after intervention (details of effect presented in Table 6)
PHCP involvement in intervention deliverySignificant outcome*
  • *

    Schwartz et al. (32) described two intervention arms that differed regarding the health professional involvement in intervention delivery; hence, these are recorded separately (see text).

  • BMI, body mass index; HOMA-IR, homeostasis model assessment of insulin resistance; LDL, low-density lipoprotein; PHCP, primary healthcare professional.

Medical professionals onlyPercentage overweight (17)
Nutrition scores (29)
Intake of snacks (32)
Both medical and allied health professionalsDining out (32)
Proportion of children with a decreased BMI z-score; and eating and physical activity behavioural skills use (18)
Weight; BMI; body fat percentage; total cholesterol; LDL; habitual physical activity index; and endurance time (22)
BMI; BMI percentile; and weight (19)
Dietary intake (11)
Sedentary behaviour decreased; physical activity increased (boys self-report), and proportion of girls meeting dietary intake guidelines (24)
BMI z-score (Group 2); systolic blood pressure (23)
Allied health professionals, after assessment and referral by medical and nursing staffLower increase in median triceps skin-fold; and fat mass percentage (14)
BMI; estimated fat mass; weight; body fat; total cholesterol; fasting insulin; and HOMA-IR (31)
Weight for height; BMI; BMI z-score; proportion of participants decreased weight-for-height (28)

Personnel training

Eight of the nine interventions that incorporated specific training for delivery personnel before the intervention commencement reported effective outcomes: five involved training for medical professionals (17,18,24,29,32), four involved training for allied health professionals (18,28,31,32) and one involved teacher training in nutrition (14).

Behaviour change

Behaviour change targets most used by effective interventions were: incorporating both healthy diet and activity into the daily routine (n = 5), decreasing sedentary behaviour (n = 4), maintaining a calorie restricted diet (n = 4), attending physical activity sessions (n = 4) and achieving a healthier diet (n = 3).

The majority of effective interventions reported employing a combination of: counselling or education (n = 11), provision of written resources (n = 11) and motivation or support (n = 9) in order to affect behaviour change. Each remaining category was associated with at least two successful interventions.


Each category of agent-of-change (child only, child and parent, parent only) was associated with effective outcomes. Higher parental involvement appeared to be incorporated into interventions targeting younger children with higher adiposity.

Number and rate of contacts and intervention delivery period

Twelve intervention arms from 11 studies are reported in this section. The two intervention arms described by Schwartz et al. (32) varied in their contact rate and are represented separately. One intervention (14) did not clearly describe the number of contacts and was excluded from this analysis.

Effective intervention arms involved between one and 114 contacts in a period of 3–12 months (Table 6). There was no indication from the literature that intervention delivery period (in months) affected intervention outcomes (even when subdivided according to intervention contact rate). However, the literature does indicate that the number of contacts has an influence. Significant anthropometric outcomes were only reported in interventions that involved five or more contacts over the intervention delivery period.

There is good evidence that contact rate intensity influences intervention outcomes (Table 6). The monthly rate of contacts in effective interventions ranged from 0.2 (one contact in 6 months) to 11.3 (34 contacts in 3 months). The six interventions with a contact rate of monthly or less (rate <2) reported one to three significant outcomes in either anthropometry (17) or behaviour change (11,24,29,32). By contrast, each of the six effective interventions with higher intensity (at least one contact every 2 weeks, rate ≥2) reported two or more significant outcomes across three domains: anthropometric (18,19,22,23,28,31), metabolic (22,23,31) and behavioural (18,22). All three interventions that reported significant metabolic outcomes had more than one contact each week, involved allied health professionals and at least one physical activity session per week.


Seventeen controlled intervention studies were identified that involved primary care in the treatment of childhood overweight and obesity; 12 of these reported significant intervention effect, when assessed using a variety of outcome measures, with no reports of adverse effects. Results of the detailed examination of these 12 interventions are synthesized in Table 8.

Table 8.  Evidence-based outcomes of this systematic review
  1. Components of effective interventions that may be considered for future treatment of childhood overweight and obesity in primary care are highlighted.

• Twelve of the 17 interventions involving primary care reported significant effect in the first post-intervention outcomes. Seven of these maintained significant effect between 6 months and 4 years after intervention.
• Effective interventions reported one or more significant anthropometric, metabolic and/or behavioural outcomes with no adverse effects.
• Effective interventions were delivered over a large range of intensities: from one contact in 6 months to three per week. The literature indicates that the contact number and intensity of interventions may be tailored according to the desired outcomes as follows:
 • Behaviour change is feasible in short, low-intensity interventions (one contact per month or less). Changes may be evident after one appointment when motivational interviewing is used.
 • Intense interventions (at least one contact per week), involving physical activity sessions and allied health, reported metabolic improvements.
 • Anthropometric changes were only reported in interventions delivered in five or more contacts.
• The evidence does not indicate that the duration of the intervention is related to outcomes.
When effectiveness is assessed using a variety of outcome measures, the evidence supports the use of the following components during interventions to treat childhood overweight and obesity in primary care.
• The provision of specific training for health professional before intervention.
• The role of medical professional in delivery of interventions as well as assessment and referral of children.
• The use of up to three of the following behaviour change targets:
 • Incorporating both healthier diet and activity into the daily routine
 • Decreasing sedentary behaviour
 • Maintaining a calorie restricted diet
 • Attending physical activity sessions
 • Achieving a healthier diet.
• The use of up to three of the following strategies to affect behaviour change:
 • Counselling or education
 • Provision of written resources
 • Motivation or support

Implications for best practice in primary care

This synthesis of the evidence indicates that the primary care setting can be effective in treating childhood overweight and obesity. This contradicts a view long-held among some health professionals that intervention in primary care to treat overweight and obese children is not effective (33–35). Removing this barrier to treatment will facilitate the implementation of interventions through primary care. In addition, the review suggests that the involvement of medical health professionals may influence intervention effect and may be expanded from assessment and referral (34) to include intervention delivery. Low-intensity interventions that were delivered solely by medical health professionals were particularly successful at affecting behaviour change.

Of the 12 effective interventions in this review, seven incorporated specific training for primary healthcare professionals. A lack of training is perceived by primary care physicians as a barrier to treating childhood overweight and obesity (36) and health professionals have reported a high level of interest in attending additional training (37). Structured training for health professionals has been successful in improving self-efficacy, attitudes and knowledge around childhood overweight and obesity (38). The evidence suggests that training before intervention delivery is an important step in primary care interventions.

Effective interventions each identified at least two targets for behaviour change, with the following being well supported: incorporating healthy diet with or without activity into the daily routine, decreasing sedentary behaviour, restricting calorie intake and attending physical activity sessions. These targets are in concordance with established guidelines for promoting healthy behaviours in primary care (39,40).

The majority of effective interventions in this review used a technique for motivation enhancement. Use of the motivational interviewing technique to support behaviour change, has been shown to outperform traditional advice giving in 80% of studies (41) and delivery in a small number of contacts is feasible in primary care (32,41). Long-term support and motivation to comply, of the type that may be provided by primary care, is likely to facilitate uptake and maintenance of desired behaviours (40,42).

There are indications from the literature that enhanced parental involvement as the sole agent-of-change improves long-term outcomes (43); however, this effect was not apparent in this review. Similarly, a recent review on the subject was not able to establish advantages of including parents (44).

This review indicates that consideration be given to the desired outcomes (anthropometric, metabolic or behavioural change) when intervention contact frequency is being determined (Table 8).

Limitations of published research

In this review, few of the included articles scored highly for MR. Insufficient data analysis, lack of an effective control group and inadequate sample size may have been responsible for apparent lack of significant outcomes, rather than efficacy of intervention components. Lower MR scores particularly reflected deficiencies in reporting and analysis. The presentation of this score in each outcome table has maintained the transparency of this review, without inappropriately excluding potentially informative studies (45).

A lack of variety in outcome measures, particularly those reflecting short-term behavioural changes, limited the ability of individual studies to assess changes that may be precursors to anthropometric and metabolic changes in the long term (4).

All interventions that reported significant behaviour changes were measured by self-report. The results of self-reported behaviour change may indicate improved knowledge of desirable behaviours rather than actual behaviour change and must be interpreted with caution due to potential bias from overreporting (46). Objective measures of behaviour change, such as accelerometry to measure physical activity/sedentary behaviour change (47) and urinary biomarkers to measure dietary compliance (46), are not subject to self-report bias.

The recruitment strategy of the majority of studies relied upon response to media advertisements; hence, participants were motivated before consent. This does not reflect intervention in primary care where eligible families are recruited prior to seeking treatment (30).

Limitations of this systematic review

The search strategy only included published literature; this may have introduced publication bias and an overrepresentation of effective interventions. Authors of included articles were not contacted for additional information. The published articles were the sole source of data; hence, the MR score may underestimate actual intervention rigour if criteria were not reported. However, other systematic reviews into treatment of childhood overweight and obesity have noted a similar paucity of interventions with high MR (4,5).

Detailed comparisons between intervention components and specific outcomes were hampered by the limited number of interventions that utilized measures from more than two domains. The inconsistency of evaluation measures between studies meant that a meta-analysis of effect size was not possible.


This systematic review has compiled the literature describing interventions to treat childhood overweight and obesity that have involved primary care. Twelve of the 17 studies reported some statistically significant intervention effect. The outcomes of this systematic review provide evidence for the use of specific intervention components (summarized in Table 8) in future intervention studies to treat childhood overweight and obesity in primary care. These intervention components are immediately practicable to future intervention studies in primary care.

Conflict of Interest Statement

No conflict of interest was declared.


The University of New South Wales (UNSW) Rural Clinical School and PhD scholarship funded by Australian Federal Government Department of Health & Ageing. Small project funding was granted by UNSW Primary Health Research Network Research Capacity Building Program. The authors thank Ms Jenny Price, librarian at Wagga Wagga Health Campus, for her assistance during the development of the database search strategy and Dr Helen Scicluna for her input during the early reviewing stages.